**** BEGIN LOGGING AT Tue Jan 12 02:59:59 2016 Jan 12 03:01:26 wyrm: are you talking server or client problem? Jan 12 07:31:05 I am trying to follow the latest document for AGL on Porter board. I could set up till bitbake step. But during bitbake, I am hitting this error. http://pastebin.com/uYaedsJZ Jan 12 07:31:07 Also, please see my local.conf file. http://pastebin.com/TD96mjpW Jan 12 07:31:11 Any help please Jan 12 07:36:26 Here is my bb file http://pastebin.com/J5hbLZdE Jan 12 08:27:13 hello Jan 12 08:27:24 any inputs would be helpful Jan 12 08:32:40 ERROR: Required build target 'agl-demo-platform' has no buildable providers. Missing or unbuildable dependency chain was: ['agl-demo-platform', '#'] Jan 12 10:02:54 hello. I am having trouble building yocto because of a github connection refused error. If I try to use the uri given in the receipe on command line (git clone) I get the same error. But the repository does exist with the uri and git clone works if I use https:// instead of git:// at the beginning Jan 12 10:04:12 ap0 for qt repositories? Jan 12 10:04:45 <_stephan> ap0 you can use git://......;protocol=https Jan 12 10:05:58 fsdun: no, it's this one: https://github.com/SolidRun/linux-imx6-3.14 Jan 12 10:06:07 _stephan: thank you. will try that right now. Jan 12 10:06:29 ah. qt was on gitorious Jan 12 10:06:48 <_stephan> http://git.yoctoproject.org/cgit.cgi/poky/plain/bitbake/lib/bb/fetch2/git.py <= details can be found here :) Jan 12 10:08:03 _stephan: thank you a lot! it atleast doesn't stop at do_fetch right away. Jan 12 10:08:42 <_stephan> you're welcome. I was a little confused by it in the beginning, too. it's quite different to git command line usage :) Jan 12 10:49:00 what's the mechanism that runs generate_index_files(d) to update the package index? i assumed something (e.g., some image bbclass) would depend on package-index.bb, but i can't find it in that case. Jan 12 11:17:55 Is someone working at wic development here? Jan 12 12:14:16 I was wondering whether there is a unique ID on each build (build name?) which could be added to each event originated by that build; is that something that exists or which has been considered? Jan 12 12:24:52 hi, if I am modify the kernel source in $WORKDIR/git, why do I get no compilation errors when I intentionally do some syntax error in the code? Jan 12 12:25:00 I tried bitbake -f -c compile linux-foo Jan 12 12:25:17 I am just trying to understand why that command comes back with no error when there is some error expected. Jan 12 12:28:40 lpapp: sure you're modifying code that's built? Jan 12 12:29:08 try the top-level makefile Jan 12 12:29:52 yeah, I am modifying such code. Jan 12 12:31:04 check if bitbake -e linux-foo | grep WORKDIR gives the directory you expect Jan 12 12:31:17 since I went to the source using that, the answer is yes :) Jan 12 12:31:26 sorry for not having made that clear. Jan 12 12:31:35 ok, dunno then :/ Jan 12 12:32:01 yeah, I had the thought of manually running make in that directory, but I do not think that is feasible for cross-building Jan 12 12:32:09 there is probably more heuristics to it. Jan 12 12:32:25 I am trying to build this for arm. Jan 12 12:51:57 is there some neat way to avoid yocto rebuilding ~everything when changing .py implementation files? Jan 12 12:52:01 for debugging purposes Jan 12 12:52:07 package_manager.py in this case Jan 12 13:23:08 Hi All, I am trying to make an image recipe that installs just one package. We have a multi-partition embedded system. At the moment I am getting a lot of other packages that i don't need, besides the one that i have. Is there a way to make such an image recipe ? Jan 12 14:04:17 Good morning! I have a quick question about PACKAGECONFIG, it seems most parameters are set using this notation: PACKAGECONFIG[udev] = "-libudev,-no-libudev,udev" Jan 12 14:04:48 or similar, what is the logic to -set -unset -set again? Jan 12 14:06:48 dotb, What exactly are you asking again? Jan 12 14:07:08 I saw a similar question raised in the mailing list https://lists.yoctoproject.org/pipermail/yocto/2013-November/016969.html Jan 12 14:08:15 I'm not sure what isn't clear Jan 12 14:08:30 Basically, I want to add the icu feature to meta-qt, PACKAGECONFIG[icu] = "-icu,-no-icu,icu" Jan 12 14:10:57 Well, syntax for packageconfig is bit inwieldly to understand at first, I'm trying to figure out how I could enable the icu feature, in the meta-qt layer. Do I just modify the meta-qt package or do I create a bbappend file? Also I saw confusion regarding ??= ?= and += Jan 12 14:11:22 make sure you need something like PACKAGECONFIG_pn_gnuradio_append = "icu" Jan 12 14:11:29 I'm hunting for exact synatx Jan 12 14:11:35 in local.conf Jan 12 14:11:41 or you can also use bbappend Jan 12 14:12:08 > in my local.conf: Jan 12 14:12:08 > PACKAGECONFIG_pn-vlc_append = " libva" Jan 12 14:13:22 Crofton|work: Thank you, testing it out now. Jan 12 14:13:59 You can use bitbake -e foo | grep ^PACKAGECONFIG to see what the var gets set to Jan 12 14:14:04 I think :) Jan 12 14:55:45 Hi all -- looking at this: http://layers.openembedded.org/layerindex/branch/master/recipes/?q=yaml Jan 12 14:56:26 i noticed that meta-python has 3.10 and some other locations have 3.11 -- what do I need to do to make sure a recipe is using 3.11? Do I need to add the custom layer in my conf.bb or make sure that the LAYERDEPENDS has the correct layer dependencies? Jan 12 14:59:04 * Crofton|work rages against layers that duplicate recipes again Jan 12 14:59:41 basically, I suggest submitting the update to meta-python and ignoring meta-ros and meta-virtualization Jan 12 15:04:21 Crofton|work: some packages in meta-virt require that version, it's hard to make everybody happy :) Jan 12 15:05:19 ok, I know meta-virt is sort of special, but they could at least submit the update to meta-python Jan 12 15:05:47 I don't see why both versions couldn't exist in meta-python. Jan 12 15:05:55 it just looks bad to see the same recipe maintained in three layers Jan 12 15:06:04 The only reason 3.11 exists, as far as I can tell, is just an update against the latest libyaml and compilers. Jan 12 15:06:32 I also sent an email a while ago about trying to get a recipe working (it has a lot of compilation errors) with bitbake, but I've gotten no helpful responses yet :( Jan 12 15:06:33 Crofton|work: true, I agree Jan 12 15:06:57 kratsg, what recipe? Jan 12 15:07:01 https://github.com/kratsg/meta-l1calo/blob/master/recipes-core/root/root_6.04.12.bb Jan 12 15:07:12 It's one that I'm trying to develop to use ROOT on an embedded board. Jan 12 15:07:27 However, I run into a lot of Fortran and CMake problems. Jan 12 15:07:34 (with C++11 support and so on) Jan 12 15:07:50 * Crofton|work curses FORTRAN :) Jan 12 15:07:53 The log is here: https://gist.github.com/kratsg/cd3184a167676eb69067 Jan 12 15:08:20 (note that this is a layer, so you can easily check it out, update an image to depend on "root" and then try to compile it yourself with bitbake :-/) Jan 12 15:08:54 and here is CMakeError.log https://gist.github.com/anonymous/adec2839deecce33702a Jan 12 15:09:00 | LLVM requires C++11 support but the '-std=c++11' flag isn't supported. Jan 12 15:09:06 look suspicious Jan 12 15:09:28 Yeah, it's really suspicious. I use CMake with this version of ROOT to get it on my computer, so I have no reason to think it would break inside bitbake :-/ Jan 12 15:09:35 [it's very recent too!] Jan 12 15:11:27 I've attempted to try and replace it with a newer version of LLVM, but not sure Jan 12 15:15:36 or think the llvm expert is asleep Jan 12 15:15:47 I think we have a couple of llvm versions you could try,m Jan 12 15:16:38 yeah, the really strange part is that ROOT does come with LLVM support internally and can use that Jan 12 15:16:43 but it tends to always default to an external one Jan 12 15:17:46 https://root.cern.ch/building-root Jan 12 16:19:21 nerdboy: Server problems. With dhcpd. Jan 12 16:30:53 Trying to compile something in Poky with "make install" and get "make: /usr/bin/install: Command not found". Any idea what's going on? Jan 12 16:31:24 does your build machine not have the install command? Jan 12 16:31:29 oh, you're compiling on target? Jan 12 16:32:03 either way, checking that /usr/bin/install exists seems like a good start Jan 12 16:32:24 /usr/bin/install comes from coreutils. if you don't have that, then your device is nowhere near equipped to do development Jan 12 16:32:58 I'm trying to get dtc installed using instructions lines 73-76 here: https://github.com/01org/edison-u-boot/blob/master/doc/README.fdt-control Jan 12 16:33:30 installing coreutils Jan 12 16:33:48 worked :) Jan 12 16:36:58 Hmm "make install" seemed to work but not "dtc: not found" Jan 12 16:37:16 1 Jan 12 16:46:36 Any idea how to install dtc library on Poky? Jan 12 16:56:39 Hello Yoctocentti - anyone know how to install device-tree-compiler on a working Poki? Jan 12 17:15:34 geotoad: Looks like you probably just install the dtc package. http://git.yoctoproject.org/cgit.cgi/poky/plain/meta/recipes-kernel/dtc/ Jan 12 17:28:45 armpit: did you see Robert's email regarding the libxml and libssl CVE patchset failures? Seems 2 of the patches might have been corrupted. Jan 12 17:30:35 wyrm: Thanks. Do you know how I can install that on a working system? Not so familiar with Yocto.. Jan 12 17:32:52 Honesly, no. I've been focusing on building things with bitbake when setting up a system, and then treating it as static afterward. Jan 12 17:33:18 Package management within a running system is not in my skillset yet. Jan 12 17:34:55 if you have package manager on the system just copy the package over and install Jan 12 17:35:15 or unpack package manually and copy files Jan 12 17:40:42 nerdboy: I only have opkg. know if I can point it to the URL or something? Jan 12 17:40:59 rburton: yt ? Jan 12 17:41:29 khem: vaguely. what's up? Jan 12 17:44:10 rburton: when is your pull request gonna land any idea ? Jan 12 17:44:23 I have phase4 lining up Jan 12 17:44:46 this will be last big one Jan 12 17:45:01 then after that it will be smaller increments Jan 12 17:45:29 opkg will work if you have an ipk Jan 12 17:45:45 or you could try this: Jan 12 17:45:49 wget -c https://github.com/hephaex/toolchain/tree/master/arm/dtc.sh Jan 12 17:46:03 make it executable and run it Jan 12 17:50:54 nerdboy: that repo looks incompatible - uses apt-get Jan 12 17:51:10 khem: next mut will have it in Jan 12 17:54:51 then you'll need to bitbake dtc and copy it over Jan 12 17:55:38 opkg can install from package feed but you'd need to have a web server serving up the feed url Jan 12 17:56:21 just scp/rsync the .ipk file and do "opkg install foo.ipk" Jan 12 18:03:19 nerdboy: thanks that's useful. Not sure where to get the dtc idk though.. Jan 12 18:03:35 * .ipk.. Jan 12 18:06:08 build it with bitbake Jan 12 18:06:30 build-imx233mini/tmp/deploy/ipk/armv5e/dtc_1.4.1+git0+302fca9f4c-r0_armv5e.ipk Jan 12 18:07:22 Ah I see. Alas I already have a build I have to work with. It's a mature project.. Jan 12 18:07:44 in that build env try "bitbake dtc" Jan 12 18:20:31 Holy cr*p this is turning out to be hard work Jan 12 18:20:49 How can it be so difficult to install a library.. Jan 12 18:27:19 So, hey, nerdboy... You asked about my dhcpd problem. So, The bitbake-built version (in Jethro) starts fine and works fine, but when I try to stop it, it fails to stop. The normal way of terminating dhcpd is to send it a sigterm, but the handler just returns and dhcpd keeps running. Jan 12 18:28:05 If I copy the sources into the running system and build it there, it starts, runs, and stops just like it's supposed to. Jan 12 18:32:39 The last build I did of Fido didn't have this behavior. So, I went through the git logs and couldn't find any likely causes of that change in behavior. Now, I'm selectively applying the package's patches trying to recreate the problem. Jan 12 18:32:57 Any other thoughts or suggestions on how to troubleshoot this? Jan 12 18:34:10 not really, i have local dnsmasq for that Jan 12 18:34:56 don't normally try to run a server on embedded stuff Jan 12 18:35:44 nerdboy: thanks for your help dude Jan 12 18:35:49 ciao Jan 12 18:53:22 I might jump ship over to DHCP subsystem of dnsmasq. I'm not quite ready to run up the white flag yet, though. Jan 12 19:02:10 wyrm: try upping the debug level on the server side? maybe run it in the foreground with strace? Jan 12 19:29:28 nerdboy: I did use strace. That's how I found that the signal handler was just returning in the broken version, instead of allowing the server to exit. Jan 12 19:30:22 The funny thing is that I can't seem to reproduce it with a hand-compiled copy. Not even with all the Poky patches applied. Jan 12 19:31:20 Also, I'm glad I had to type that insead of saying it. "Poky patches applied" seems like a good tongue twister. Jan 12 19:37:21 wyrm: you didn't get any configure warnings did you? Jan 12 19:37:40 try comparing config.log from the two builds? Jan 12 20:04:46 Well, if I hand-compile with --with-libbind (which the bitbake recipe uses), ./configure finishes fine but make fails. Jan 12 20:04:49 | In file included from ../includes/dhcpd.h:91:0, Jan 12 20:04:51 | from alloc.c:30: Jan 12 20:04:54 | ../includes/omapip/isclib.h:51:24: fatal error: isc/buffer.h: No such file or directory Jan 12 20:05:49 Without --with-libbind, ./configure and make succeed, and dhcpd runs and stops correctly. Jan 12 20:06:37 However! Since you mentioned it, I looked, and yes. The hand-compiled version does produce an error when it tries to build conftest.c Jan 12 20:07:11 That doesn't seem to impair the build process, though. Jan 12 20:07:50 * zeddii has no context. but wyrm .. what branch are you using ? and what was the target ? Jan 12 20:09:23 are any of these headers even kernel headers? Jan 12 20:11:45 zeddii: Upgrading from Fido to Jethro. The target is ... fsl-multimedia-image-full. Jan 12 20:12:50 The package in question is dhcp. Under fido, dhcpd-4.3.1 built, ran, and stopped correctly. Under jethro, dhpcd-4.3.2 builds and runs, but won't stop on a SIGTERM. Jan 12 20:13:55 I can hand-compile dhcpd-4.3.2 and it will build, run, and stop correctly. I can't seem to re-create the error by hand. Jan 12 20:14:48 it is worth checking to see if prelink is in play. I've had it corrupt my binaries in the past (but it should all be fixed in master), which meant that when I built by hand, everything worked. Jan 12 20:16:10 I just used a bbappend to remove the --with-libbind option, and I'm trying to eliminate the error that way (since that was the one thing I couldn't do by hand). Jan 12 20:16:37 But prelinking works perfectly all the time! Jan 12 20:16:38 ... Jan 12 20:16:41 I'll look. Jan 12 20:23:26 definitely rebuild w/ prelinking off.. (disabling prelink only affects the image generation, so it's quick trial process) Jan 12 20:23:34 if you can find a reproducible problem, please file a bug and let me know.. Jan 12 20:23:49 otherwise, I'd say it's likely "something else", some kind of contamination in the build Jan 12 20:26:32 Yep. Prelinking is enabled. I'll disable that and try again once it's done doing the build it's doing now. Jan 12 20:27:03 (you can also unprelink the one binary, on the target, if you have the target prelink recipe included in your image) Jan 12 20:29:13 unprelink? That sounds like an automatic language fine of $2 right there.... :) Jan 12 20:36:31 Nope. I don't. Jan 12 20:36:55 So, I rebuild. Jan 12 21:05:48 does anyone know why an autotools-inheriting recipe would have oe_runconf fail but the configure exit 0? Jan 12 21:42:41 meh Jan 12 21:42:53 updating U-Boot to 2016.01 produces some weird problems Jan 12 21:43:14 can someone please explain to me what does EXTRA_OEMAKE_class-target and EXTRA_OEMAKE_class-cross mean ? Jan 12 21:43:40 more like what is the difference between "target" and "cross" Jan 12 22:27:36 Marex: cross = built to run on the build machine, but target the target. target builds binaries that run on target, not the build machine. the main example of a cross tool is the crosscompiler. Jan 12 22:28:12 kergoth: thanks Jan 12 22:28:22 kergoth: now it makes sense to me Jan 12 23:33:31 kergoth: nativesdk, crosssdk, cross-canadian, native also need to be explained ;) Jan 13 02:33:35 * nerdboy has never met a cross canadian **** ENDING LOGGING AT Wed Jan 13 02:59:59 2016